Plasminogen activator inhibitor-1 (PAI-1) is a 50 kDa single-chain glycoprotein (Loskutoff et al., Proc. Natl. Acad. Sci. USA 80:2956-2960, 1983; Chmielewska et al., Thromb. Res. 31:427-436, 1983) that is the principal inhibitor of both urokinase type plasminogen activator (uPA) and tissue type PA (tPA) (Fay et al., N. Engl. J. Med. 327:1729-1733, 1992; Fay et al., Blood 83:351-356, 1994; Yepes et al., “Plasminogen Activator Inhibitor-1,” In Hemostasis and Thrombosis: Basic Principles and Clinical Practice, Robert W Colman et al., editors Lippincott Williams & Wilkins, pp. 365-380, 2006). PAI-1 inhibits tPA and uPA with second-order rate constants ˜107 M−1s−1, a value that is 10-1000 times faster than the rates of PA inhibition by other PAIs (Sprengers et al., Blood 69:381-387, 1987; Lawrence et al., J. Biol. Chem. 265:20293-20301, 1990; Lawrence et al., In Molecular Basis of Thrombosis and Hemostasis. K. A. High, and Roberts, H. R., editors. Marcel Dekker, Inc. New York. 517-543, 1995). Moreover, approximately 70% of the total tPA in carefully collected normal human plasma is detected in complex with PAI-1, suggesting that inhibition of tPA by PAI-1 is a normal, ongoing process. PAI-1 can also directly inhibit plasmin (Hekman et al., Biochem. 27:2911-2918, 1988; Stefansson et al., J. Biol. Chem. 276:8135-8141, 2001). Thus, PAI-1 is the chief regulator of plasmin generation in vivo, and as such it appears to play an important role in both fibrinolytic and thrombotic disease (Booth, Baillieres Best. Pract. Res. Clin. Haematol. 12:423-433, 1999; Huber, J. Thromb. Thrombolysis. 11:183-193, 2001; Kohler et al., N. Engl. J. Med. 342:1792-1801, 2000; Stefansson et al., Current Pharmaceutical Design 9:1545-1564, 2003). PAI-1 has three potential N-linked glycosylation sites and contains between 15 and 20% carbohydrate (Ny et al., Proc. Natl. Acad. Sci. USA 83:6776-6780, 1986; van Mourik et al., J. Biol. Chem. 259:14914-14921, 1984).
PAI-1 belongs to the Serine Protease Inhibitor super family (SERPIN), which is a gene family that includes many of the protease inhibitors found in blood, as well as other proteins with unrelated or unknown functions (Gettins et al., Serpins: Structure, Function and Biology. R.G. Landes Company. Austin, Tex. U.S.A., 1996). Serpins are consumed in the process of protease inactivation and thus act as “suicide inhibitors” (Lawrence et al., J. Biol. Chem. 270:25309-25312, 1995). The association between a serpin and its target protease occurs at an amino acid residue, referred to as the “bait” residue, located on a surface loop of the serpin called the reactive center loop (RCL) (Huber et al., Biochem. 28:8951-8966, 1989; Sherman et al., J. Biol. Chem. 267:7588-7595, 1992). The “bait” residue is also called the P1 residue, and is thought to mimic the normal substrate of the enzyme. Upon association of the P1 residue with the S1 site of a target protease, cleavage of the RCL occurs. This is coupled to a large conformational change in the serpin which involves rapid insertion of the RCL into the major structural feature of a serpin, β-sheet A (Lawrence et al., J. Biol. Chem. 270:25309-25312, 1995; Lawrence et al., J. Biol. Chem. 270:25309-25312, 1995; Wilczynska et al., J. Biol. Chem. 270:29652-29655, 1995; Lawrence et al., J Biol. Chem. 275:5839-5844, 2000; Hagglof et al., J. Mol. Biol. 335:823-832, 2004). This results in tight docking of the protease to the serpin surface and to distortion of the enzyme structure, including its active site. RCL insertion also produces a large increase in serpin structural stability making the complex rigid and thus trapping the protease in a covalent acyl-enzyme complex with the serpin (Lawrence et al., J. Biol. Chem. 265:20293-20301, 1990; Huntington et al., Nature 407:923-926, 2000; Huntington et al., Sci. Prog. 84:125-136, 2001).
Native PAI-1 exists in at least two distinct conformations, an active form that is produced by cells and secreted, and an inactive or latent form that accumulates in cell culture medium over time (Hekman et al., J. Biol. Chem. 260:11581-11587, 1985: Mottonen et al., Nature 355:270-273, 1992). In blood and tissues, most of the PAI-1 is in the active form; however, in platelets both active and latent forms of PAI-1 are found (Erickson et al., J. Clin. Invest. 74:1465-1472, 1984). In active PAI-1, the RCL is exposed on the surface of the molecule (Sharp et al., Structure Fold. Des. 7:111-118, 1999), but upon reaction with a protease, the cleaved RCL integrates into the center of β sheet A (Lawrence et al., J Biol. Chem. 270: 25309-25312, 1995; Sharp et al., supra). In the latent form, the RCL is intact, but instead of being exposed, the entire amino terminal side of the RCL is inserted as the central strand into the β sheet A (Mottonen et al., supra). This accounts for the increased stability of latent PAI-1 as well as its lack of inhibitory activity (Hekman et al., supra; Lawrence et al., Biochem. 33: 3643-3648, 1994; Lawrence et al., J. Biol. Chem. 269: 27657-27662, 1994).
Active PAI-1 spontaneously converts to the latent form with a half-life of one to two hours at 37° C. (Hekman et al., Biochem. 27:2911-2918, 1988; Lawrence et al., Eur. J. Biochem. 186:523-533, 1989; Levin et al., Blood 70:1090-1098, 1987; Lindahl et al., Thromb. Haemost. 62:748-751, 1989), and latent PAI-1 can be converted back into the active form by treatment with denaturants (Hekman et al., J. Biol. Chem. 260:11581-11587, 1985). Negatively charged phospholipids can also convert latent PAI-1 to the active form, suggesting that cell surfaces may modulate PAI-1 activity (Lambers et al., J. Biol. Chem. 262:17492-17496, 1987). The observation that latent PAI-1 infused into rabbits is apparently converted to the active form is consistent with this hypothesis (Vaughan et al., Circ. Res. 67:1281-1286, 1990). The spontaneous reversible interconversion between the active and latent structures is unique for PAI-1 and distinguishes it from other serpins; however, the biological significance of the latent conformation remains unknown.
Other non-inhibitory forms of PAI-1 have also been identified. The first form results from oxidation of one or more critical methionine residues within active PAI-1 (Lawrence et al., Biochem. 25:6351-6355, 1986; Strandberg et al., J. Biol. Chem. 266:13852-13858, 1991). This form differs from latent PAI-1 in that it can be partially reactivated by an enzyme that specifically reduces oxidized methionine residues (Lawrence et al., Biochem. 25:6351-6355, 1986). Oxidative inactivation of PAI-1 may be an additional mechanism for the regulation of PAI-1, and oxygen radicals produced locally by neutrophils or other cells may inactivate PAI-1 and thus facilitate the generation of plasmin at sites of infection or in areas of tissue remodeling (Weiss et al., J. Clin. Invest. 73:1297-1303, 1984). PAI-1 also exists in two different cleaved forms. As noted above, PAI-1 in complex with a protease is cleaved in its RCL. Uncomplexed PAI-1 can also be found with its RCL cleaved, which can arise from dissociation of PAI-1-PA complexes or from cleavage of the RCL by a non-target protease at a site other than the P1 (Lawrence et al., J Biol. Chem. 270:25309-25312, 1995; Lawrence et al., J. Biol. Chem. 269:27657-27662, 1994; Wu et al, Blood 86:1056-1061, 1995). None of these forms of PAI-1 are able to inhibit protease activity; however, they may interact with other ligands.
The interaction of PAI-1 with non-protease ligands plays an essential role in PAI-1 function (Yepes et al., supra). PAI-1 binds with high affinity to heparin, the cell adhesion protein vitronectin, and members the endocytic low-density lipoprotein receptor (LDL-R) family, such as the lipoprotein receptor-related protein (LRP), and the very low density lipoprotein receptor (VLDL-R) (Lawrence et al., J Biol. Chem. 269:15223-15228, 1994; Stefansson et al., J Biol. Chem. 273:6358-6366, 1998; Horn et al., Thromb. Haemost. 80:822-828, 1998; Hussain et al., Annu. Rev. Nutr. 19:141-172, 1999; Jensen et al., FEBS Lett. 521:91-94, 2002; Zhou et al., Nat. Struct. Biol. 10:541-544, 2003; Xu et al., J. Biol. Chem. 279:17914-17920, 2004). These non-protease interactions are important for both PAI-1 localization and function, and they are largely conformationally controlled through structural changes associated with RCL insertion (Seiffert et al., J. Biol. Chem. 272:13705-13710, 1997; Podor et al., J. Biol. Chem. 275:19788-19794, 2000; Webb et al., J Cell Biol 152:741-752, 2001; Minor et al., J. Biol. Chem. 277:10337-10345, 2002; Stefansson et al., Sci. STKE. 2003:e24, 2003; Stefansson et al., J. Biol. Chem. 279: 29981-29987, 2004). In blood, most of the active PAI-1 circulates in complex with the glycoprotein vitronectin. The PAI-1 binding site for vitronectin has been localized to a region on the edge of β-sheet A in the PAI-1 structure (Lawrence et al., J Biol. Chem. 269:15223-15228, 1994; Jensen et al., FEBS Lett. 521:91-94, 2002; Zhou et al., Nat. Struct. Biol. 10:541-544, 2003; Xu et al., J. Biol. Chem. 279:17914-17920, 2004). The binding site for LDL-R family members is less well characterized, but has been identified, in a region of PAI-1 associated with alpha helix D that is adjacent to the vitronectin binding domain (Stefansson et al., J Biol. Chem. 273:6358-6366, 1998; Horn et al., Thromb. Haemost. 80:822-828, 1998). The heparin binding domain on PAI-1 has also been mapped. This site also localizes to alpha helix D in a region homologous to the heparin binding domain of antithrombin III (Ehrlich et al., J. Biol. Chem. 267:11606-11611, 1992), and may overlap with the binding site for LDL-R family members.
Vitronectin circulates in plasma and is present in the extracellular matrix primarily at sites of injury or remodeling (Podor et al., J. Biol. Chem. 275:19788-19794, 2000; Tomasini et al., Vitronectin. Prog. Hemost. Thromb. 10:269-305, 1991; Seiffert, Histol. Histopathol. 12:787-797, 1997; Podor et al., J. Biol. Chem. 275:25402-25410, 2000; Podor et al., J. Biol. Chem. 277:7520-7528, 2002). PAI-1 and vitronectin appear to have a significant functional interdependence. Vitronectin stabilizes PAI-1 in its active conformation, thereby increasing its biological half-life (Lindahl et al., Thromb. Haemost. 62:748-751, 1989).
Vitronectin also enhances PAI-1 inhibitory efficiency for thrombin approximately 300-fold (Keijer et al., Blood 78:1254-1261, 1991; Naski et al., J. Biol. Chem. 268:12367-12372, 1993). In turn, PAI-1 binding to vitronectin alters its conformation from the native plasma form, which does not support cell adhesion, to an “activated” form that is competent to bind integrins. However, integrin binding is blocked by the presence of PAI-1 (Seiffert et al., J. Biol. Chem. 272:13705-13710, 1997). As noted above, the association of PAI-1 with vitronectin is conformationally controlled and upon inhibition of a protease, the conformational change in PAI-1 associated with RCL insertion results in a loss of high affinity for vitronectin and a gain in affinity for LDL-R family members (Stefansson et al., J Biol. Chem. 273:6358-6366, 1998; Lawrence et al., J. Biol. Chem. 272:7676-7680, 1997). This is due to RCL insertion in PAI-1, disrupting the vitronectin binding site, while simultaneously exposing a cryptic receptor binding site that is revealed only when PAI-1 is in a complex with a protease (Sharp et al., Structure Fold. Des. 7:111-118, 1999; Stefansson et al., J Biol. Chem. 273:6358-6366, 1998; Stefansson et al., J. Biol. Chem. 279: 29981-29987, 2004), which results in an approximately 100,000-fold shift in the relative affinity of PAI-1 from vitronectin to LDL-R family members and a subsequent shift in PAI-1 localization from vitronectin to the cellular receptor (Stefansson et al., J Biol. Chem. 273:6358-6366, 1998; Lawrence et al., J. Biol. Chem. 272:7676-7680, 1997). Thus, PAI-1 association with vitronectin and LDL-R is conformationally controlled.
High PAI-1 levels are associated with both acute diseases such as sepsis and myocardial infarction (Colucci et al., J. Clin. Invest 75:818-824, 1985; Hamsten et al., N. Engl. J. Med. 313:1557-1563, 1985), and with chronic disorders including cancer, atherosclerosis, and type 2 diabetes (Stefansson et al., Current Pharmaceutical Design 9:1545-1564, 2003; De et al., Curr. Opin. Pharmacol. 5:149-154, 2005; Kannel, Lipids 40:1215-1220, 2005; Durand et al., Thromb. Haemost. 91:438-449, 2004). The association of PAI-1 with these diseases or syndromes has led to the hypothesis that PAI-1 is involved in their pathology. However, the mechanistic role that PAI-1 plays in disease development is not clear and is likely to be complex since PAI-1 can act through multiple pathways, such as modulating fibrinolysis through the regulation of plasminogen activators, or by influencing tissue remodeling through the direct regulation of cell migration (Stefansson et al., Sci. STKE. 2003:e24, 2003; Stefansson et al., Nature 383:441-443, 1996; Deng et al., J Cell Biol. 134:1563-1571, 1996; Czekay et al., J. Cell Biol. 160:781-791, 2003; Cao et al., EMBO J. 25:1860-1870, 2006).
In cardiovascular disease, PAI-1 expression is significantly increased in severely atherosclerotic vessels (Schneiderman et al., Proc. Natl. Acad. Sci. USA 89:6998-7002, 1992; Lupu et al., Arterioscler. Thromb. 13:1090-1100, 1993), and PAI-1 protein levels rise consistently during disease progression from normal vessels to fatty streaks to atherosclerotic plaques (Robbie et al., Arterioscler. Thromb. Vasc. Biol. 16:539-545, 1996). Increased PAI-1 levels are also linked to obesity, and insulin resistance (Alessi et al., Diabetes 46:860-867, 1997; Loskutoff et al., Arterioscler. Thromb. Vasc. Biol. 18:1-6, 1998; Schafer et al., FASEB J. 15:18404842, 2001).
In addition, elevated plasma levels of PAI-1 have been associated with thrombotic events (Krishnamurti, Blood 69: 798, 1987); Reilly, Arteriosclerosis and Thrombosis, 11:1276, 1991), and antibody neutralization of PAI-1 activity resulted in promotion of endogenous thrombolysis and reperfusion (Biemond, Circulation 91:1175, 1995; Levi, Circulation 85: 305, 1992). Elevated levels of PAI-1 have also been implicated in polycystic ovary syndrome (Nordt, J. Clin. Endocrin. Metabol. 85:1563, 2000) and bone loss induced by estrogen deficiency (Daci, J. Bone Min. Res. 15:1510, 2000).
PAI-1 is synthesized in both murine and human adipocytes (Alessi et al., Diabetes 46:860-867, 1997; Samad et al., J. Clin. Invest. 97:37-46, 1996; Morange et al., Diabetes 48:890-895, 1999; Sakamoto et al., Am. J. Physiol. 276:C1391-C1397, 1999; Samad et al., Ann. N. Y. Acad. Sci. 811:350-358, 1997; Lundgren et al., Circulation 93:106-110, 1996; Cigolini et al., Atherosclerosis 143:81-90, 1999; Crandall et al., Biochem. Biophys. Res. Commun. 279:984-988, 2000; Gottschling-Zeller et al., Diabetologia 43:377-383, 2000). There is also a strong correlation between the amount of visceral fat and plasma levels of PAI-1 in humans (Alessi et al., Diabetes 46:860-867, 1997; Vague et al., Metabolism 35:250-253, 1986; Mavri et al., Arterioscler. Thromb. Vasc. Biol. 19:1582-1587, 1999; Giltay et al., Arterioscler. Thromb. Vasc. Biol. 18:1716-1722, 1998) and mice (Samad et al., Mol. Med. 2:568-582, 1996; Shimomura et al., Nat. Med. 2:800-803, 1996). This dramatic up-regulation of PAI-1 in obesity has lead to the suggestion that adipose tissue itself may directly contribute to elevated systemic PAI-1, which in-turn increases the probability of vascular disease through increased thrombosis, and accelerated atherosclerosis. Notably, very recent data suggests that PAI-1 may also play a direct role in obesity (Schafer et al., FASEB J. 15:1840-1842, 2001; Ma et al, Diabetes 53:336-346, 2004; De Taeye et al., J. Biol. Chem. 281: 32796-32805, 2006; Crandall et al., Arterioscler. Thromb. Vasc. Biol. 26: 2209-2215, 2006; Liang et al., Am. J. Physiol. Endocrinol. Metab. 290:E103-E113, 2006).
In one study, genetically obese and diabetic ob/ob mice crossed into a PAI-1 deficient background had significantly reduced body weight and improved metabolic profiles compared to ob/ob mice with PAI-1 (Schafer et al., FASEB J. 15:1840-1842, 2001). Likewise, nutritionally-induced obesity and insulin resistance were dramatically attenuated in mice genetically deficient in PAI-1 (Ma et al, Diabetes 53:336-346, 2004; De Taeye et al., J. Biol. Chem. 281:32796-32805, 2006) and in mice treated with an orally active PAI-1 inhibitor (Crandall et al., Arterioscler. Thromb. Vasc. Biol. 26: 2209-2215, 2006). The improved adiposity and insulin resistance in PAI-1-deficient mice may be related to the observation that PAI-1 deficient mice on a high fat diet had increased metabolic rates and total energy expenditure compared to wild-type mice, and peroxysome proliferator-activated receptor (PPARγ) and adiponectin were maintained (Ma et al, Diabetes 53:336-346, 2004). However, the precise mechanism involved was not shown and may be complex, since the over-expression of PAI-1 in mice also impaired adipose tissue formation (Lijnen et al., J. Thromb. Haemost. 3:1174-1179, 2005). Taken together, these observations suggest that PAI-1 plays a previously unrecognized direct role in obesity and insulin resistance that involves interactions beyond its identified activities of modulating fibrinolysis and tissue remodeling.
Indeed, if PAI-1 positively regulates adipose tissue development, then the association of increased PAI-1 expression with developing obesity may constitute a positive feedback loop promoting adipose tissue expansion and dysregulation of normal cholesterol homeostasis. Thus, there exists a need in the art for a greater understanding of how PAI-1 is involved in metabolism, obesity and insulin resistance. The invention provides methods of identifying and using inhibitors of PAI-1.